Latent Class and Profile Analysis: Applications to Social Work Research

In recent years, methods of analyzing latent variables have become increasingly popular in social work research. Yet, relatively few social work researchers are using person-centered methods of analyzing latent variables such as latent class analysis (LCA) or latent profile analysis (LPA). A keyword search of Social Work Abstracts (in April 2009, using the terms “latent AND [class OR cluster OR profile OR structure]”) and a hand search of five social work journals (Journal of Social Service Research, Journal of Sociology and Social Welfare, Social Service Review, Social Work Research, and Research on Social Work Practice) revealed only six social work articles using LC models (Bowen, Lee, & Weller, 2007; Keller, Cusick, & Courtney, 2007 Neely-Barnes, Marcenko, & Weber, 2008; Pears, Kim, & Fisher, 2008; Travis & Combs-Orme, 2007; Vaughn, Shook, & McMillen, 2008). Yet, these techniques have much to offer us as social work researchers.

LCA and LPA offer the opportunity to examine heterogeneity in populations of individuals, families, or communities that cannot be measured through observed variables (McCutcheon, 1987). “LCA” is the term usually used to describe the use of cross-sectional data with dichotomous or categorical measured indicators (McCutcheon, 1987a), and “LPA” describes the use of cross-sectional data with continuous indicators (Muthén, 2001). The presence of unobserved latent classes has important implications for social work research in that we may identify latent classes of individuals, families, or communities with unique needs for intervention or unique responses to intervention. As such, LCA and LPA could be seen as an emerging horizon for social work research.

LCA and LPA offer an advantage over cluster analysis in that they are model-based, the choice of cluster criteria are less arbitrary, and no decisions need to be made about scaling of observed variables (Vermunt & Magidson, 2002). Although latent class models have been described for many years (Green, 1951, 1952), recent developments of more efficient and usable statistical methods have made the application of latent class models to social science problems a more realistic possibility.

This workshop aims to introduce and teach the techniques of LCA and LPA to an audience of social work researchers. Participants will learn the underlying assumptions of LCA and LPA and the difference between exploratory and confirmatory analysis. The advantages and disadvantages of different tests of model fit and quality of classification indicators will be discussed. Participants will see demonstrations of LCA or LPA using the latest version of Mplus, but other statistical software packages will also be discussed. Step-by-step instructions on how to run statistical software packages including suggestions on writing syntax and how to interpret results will be given. Participants can expect to leave the workshop able to formulate a research question applying LCA or LPA to their own area of interest and with the skills to carry out their analyses. Finally, participants will learn how to write up the results of their analysis for publication.